Metropolis algorithm driven factor analysis for lithological characterization of shallow marine sediments

Abstract

Factor analysis of well logging data can be effectively applied to calculate shale volume in hydrocarbon formations. A global optimization approach is developed to improve the result of traditional factor analysis by reducing the misfit between the observed well logs and theoretical data calculated by the factor model. Formation shaliness is directly calculated from the factor scores by a nonlinear regression relation, which is consistent in the studied area in Alaska, USA. The added advantage of the implementation of the Simulated Annealing method is the estimation of the theoretical values of nuclear, sonic, electrical as well as caliper well-logging data. The results of globally optimized factor analysis are compared and verified by independent estimates of self-potential logbased deterministic modeling. The suggested method is tested in two different shaly-sand formations in the North Aleutian Basin of Alaska and the comparative study shows that the assumed nonlinear connection between the factor scores and shale volume is applicable with the same regression constants in different burial depths. The study shows that factor analysis solved by the random search technique provides an independent in situ estimate to shale content along arbitrary depth intervals of a borehole, which may improve the geological model of the hydrocarbon structure in the investigated area.

A combined estimator using TEC and b-value for large earthquake prediction

“…Pardalos and Romeijn 2002) and has been extensively used in the last years in the field of geosciences (e.g. Abordán and Szabó 2018, Baselga 2018, Biswas and Sharma 2017. Also worth mentioning is the fact that in order for the estimator to be as representative as possible we made use of hundreds of randomly selected days.…”